Top 10 in Tech - What to know for Week ending February 20, 2026

1. SaaS METRIC OF THE WEEK: SaaS METRIC OF THE WEEK: Revenue & Burn per Employee. In this profits-over-growth new SaaS world, forget vanity metrics. This post makes the case for tracking Revenue and Burn per Employee as core performance metrics. Why? They reveal whether you’re scaling efficiently—or just adding bodies. It also includes great benchmarks across stages.

2. DUNNING: Time for your annual reminder of the real term with a weird-ass name - it's a phrase for involuntary churn (aka bad or failed payments). According to Baremetrics SaaS and subscription businesses lose around 9% of their MRR due to failed payments on average. Learn more about a successful dunning (and pre-dunning) process.

3. UNDERSELL: If expansion fits into your growth strategy (it should) take a read of two-part series from Tomasz Tunguz and  Bill Binch - part one is deliberately underselling as a sales strategy to minimize churn and increase upsell/expansion opportunities as a land and expand strategy and post 2 is an expansion of land and expand witch details how to structure a Startup sales team for optimal land & expand.

4. TEAM: Check this stat: VCs say startup success depends on your TEAM DYNAMICS (56% of them), then timing (12%) and tech (9%). The tech sector likes teams more (64%) than healthcare (42%).

5. SAAS IS DEAD? 70% of public SaaS now trades below 5x forward multiples, down from (an admittedly bonkers and frothy) 25x+ in 2021. The market isn’t killing SaaS, but it’s repricing hype, or moving the hype around a little. According to Blackbird Ventures (and please scroll, as Blackbird has an annoying full-page banner image that makes the page look blank), “AI vs SaaS” is the wrong debate. Winners will be those who reinvent distribution and workflow leverage, not those relabeling decks. I guess we all get valued on our relevance?

6. AI MARGIN: The article above in #5 got me thinking. One of the investable things about SaaS is the margin - it's high. And AI is an expensive endeavor to do well. So I did a bit of sleuthing. SaaS built its legend on 75–90% gross margins. AI often runs 40–60% because of the tokens and compute-based architecture. The SaaSCFO did the maths, and their model showed AI needs ~6x the revenue to match SaaS EBITDA at the same cost structure. The math hasn't necessarily changed - the inputs have. TAM and ARPA have to now carry what margin no longer does.

7. PRICING: Great new report from ChartMogul, Seat-based pricing still dominates SaaS revenue, but their latest data shows per-seat plans drive the majority of ARR across B2B, with usage-based models growing but not replacing seats. Hybrid pricing is rising fast - especially in mid-market and enterprise - blending predictability with expansion upside.

8. CUSTOMER SERVICE: This is a short, but pretty interesting post - sure, tech is great at creating jobs that don't exist yet, but dang! Look at the first chart in this a16z post showing the plunge in hiring of CS roles from Q4 '23 to Q3 '25. It's about 1/3 of what it was 2 years ago.

9. BUILD vs BUY: It's a great Tech-ism and debate - but has it changed much, given the Agent/Vibe-based ease with which some tools can be built internally? AI made building ridiculously easier. BUUUUUUT - It didn’t make owning software easier. Sure, your Vibe coding lowers the cost to prototype, but not the cost to maintain, secure, audit, and evolve, etc. So........ build if it’s core differentiation. Buy if it’s your plumbing. TL;DR: AI can change speed but not your TCO.

10. CASE STUDY: In a down-round exit - Brex was acquired for $5.15B after a $12.3B peak valuation. In today's current environment, is that a bad thing? Early investors likely did great. Late-stage investors and many employee option holders? Not so much. Reminder for us all: valuation is a debt, not a trophy. Entry price and cap table structure determine a lot about who the winners are in liquidity events.

POD OF THE WEEK: Adding onto #5 and #6 above - this podcast looks at inference costs (stuff like GPU/TPU compute time, energy, and infrastructure overhead for every user prompt) with AI companies, noting that companies like Cursor and Lovable treat compute as their primary growth investment, not their primary margin drag